[DLMED] implement intensity normalizer

Normalize input based on provided args, using calculated mean and std if not provided.

monai/data/transforms/intensity_normalizer.py

@@ -0,0 +1,59 @@
+# Copyright 2020 MONAI Consortium
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#     http://www.apache.org/licenses/LICENSE-2.0
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import numpy as np
+from .multi_format_transformer import MultiFormatTransformer
+from .shape_format import get_shape_format
+from .shape_format import get_channel_axis
+
+
+class IntensityNormalizer(MultiFormatTransformer):
+    """Normalize input based on provided args, using calculated mean and std if not provided
+    (shape of subtrahend and divisor must match. if 0, entire volume uses same subtrahend and
+     divisor, otherwise the shape can have dimension 1 for channels).
+
+    Args:
+        img - the MedicalImage to be processed
+        subtrahend (ndarray): the amount to subtract by (usually the mean)
+        divisor (ndarray): the amount to divide by (usually the standard deviation)
+    """
+
+    def __init__(self, dtype=np.float32):
+        MultiFormatTransformer.__init__(self)
+        self._dtype = dtype
+
+    def _handle_any(self, img, subtrahend=None, divisor=None):
+        if subtrahend is not None and divisor is not None:
+            assert isinstance(subtrahend, np.ndarray)
+            assert isinstance(divisor, np.ndarray)
+            if subtrahend.ndim == 0 and divisor.ndim == 0:
+                img -= subtrahend
+                img /= divisor
+            else:  # we have array or matrix: current implementation, just handle array for channels
+                shape_format = get_shape_format(img)
+                assert shape_format is not None, 'can not support this shape format.'
+                channel_axis = get_channel_axis(shape_format)
+                assert len(subtrahend.shape) == 1
+                assert len(divisor.shape) == 1
+                assert subtrahend.shape[0] == img.shape[channel_axis]
+                assert divisor.shape[0] == img.shape[channel_axis]
+                img = np.moveaxis(img, channel_axis, -1)
+                img -= subtrahend
+                img /= divisor
+                img = np.moveaxis(img, -1, channel_axis)
+        else:
+            img -= np.mean(img)
+            img /= np.std(img)
+
+        if self._dtype != img.dtype:
+            img = img.astype(self._dtype)
+
+        return img

monai/data/transforms/shape_format.py

@@ -43,3 +43,19 @@ def get_shape_format(img: np.ndarray):
         return ShapeFormat.CHWD
     else:
         return None
+
+
+def get_channel_axis(fmt):
+    """Get the channel axis number
+
+    Args:
+        fmt: a shape format to analyze channel information
+
+    Returns: the channel axis if the format is channeled, or None if not.
+
+    """
+    assert type(fmt) == str, 'format must be string.'
+    for i in range(len(fmt)):
+        if fmt[i] == 'C':
+            return i
+    return None